The Australian Taxation Office (ATO) is contemplating the implementation of artificial intelligence to aid its cadre of 800 core developers in tackling prevalent programming challenges.
The envisioned AI coding assistant is expected to deliver features such as intelligent code suggestions, bug remediation, automated generation of test cases and scripts, as well as the capability to refactor codebases across diverse technological platforms.
According to a request for tender, the objective is to enable developers to concentrate on higher-value tasks, including test case strategizing, application security protocols, and the maintenance of existing technologies.
“The ATO operates within a complex, multi-faceted IT delivery landscape that supports over 800 core developers across a variety of technologies,” the agency articulated in its requirements statement.
Specifically, the ATO is pursuing a software-as-a-service solution that seamlessly integrates with Microsoft’s development environments, specifically Visual Studio 2019 and 2022, along with the Visual Studio Code editor.
This solution must also connect with the agency’s Azure DevOps and Git repositories, while managing the translation of legacy code—such as COBOL—into contemporary programming languages.
A critical security aspect emphasized in the tender stipulates that any code processed by the AI will not be stored or utilized for training purposes, thereby addressing privacy and compliance issues.
This initiative to embrace AI in software development aligns with the ATO’s larger strategy to incorporate artificial intelligence throughout its operational framework.
The agency has identified five enterprise-level use cases for AI that include fraud detection, client risk profiling, and document comprehension.
During the recent AI Innovation Showcase in Canberra, ATO Assistant Commissioner for Data Science, Ying Yang, elaborated on the agency’s trial of large multimodal AI models designed to assist in the auditing of documents submitted by taxpayers.
Yang outlined an internal framework for categorizing AI capabilities into three distinct tiers: scaled machine learning, augmentation, and intelligent automation.
She characterized scaled machine learning as systems that “mimic” human instruction, akin to a student acquiring knowledge from an educator.
The augmentation level encompasses generative AI, transitioning towards self-learning and developmental attributes.
“Humans and AI become collaborative partners, each enhancing the other’s unique strengths,” Yang stated.
The ultimate tier, intelligent automation, involves AI transcending procedural tasks to offer support in judgment, creativity, and interpersonal interactions, as delineated in Yang’s accompanying PowerPoint presentation.
Source link: Itnews.com.au.
